Find out how ICT can support biomedical and clinical researchFind out more. From Clever cars to clever farms... Embedded Systems
project description
Data mining, or knowledge discovery, is the computer-assisted process of digging through and analysing enormous sets of data and then extracting the meaning of the data. Data mining tools predict behaviours and future trends. They search databases for hidden patterns, finding predictive information that experts may miss because it lies outside their expectations. Data mining requires either sifting through an immense amount of material, or intelligently probing it to find where the value resides.


In the figure on the right you can see a map showing weekly location data for all the caribou tracked in the Porcupine Caribou Herd Satellite Collar Project. 		Map with caribou locations

In this project we will address spatio-temporal pattern discovery

Interesting patterns for moving objects involve some subset of the objects that have the same behaviour.
This can be meeting in some region, or moving along similar paths. Two simple examples are:
  • Encounter: A large enough subset of points meets in the same circular region at the same time
  • Flocking: A large enough subset of points is moving along paths close to each other for a certain (pre-defined) time
Depending on the application and the pattern, different answers might be sought. For example,
in some applications one might only want to detect whether a pattern occurs or not, while in other applications
all such patterns should be reported. Also, in some applications repetitive patterns (commuting patterns or driving patterns)
are of interest, such as:
  • Recurrence: The same behaviour is repeated many times
  • Concurrent recurrence: Same, but the behaviours must be at the same time for the subset
  • Regular recurrence: Same as recurrence, but the time interval between consecutive visits is regular

Tools

The most crucial step is the development of efficient algorithms for detecting and reporting patterns.
Since the spatio-temporal patterns are very complex compared to patterns that have been considered before,
it is unlikely that any approach neglecting the spatial information would be able to generate successful algorithms.
We believe that computational geometry will play a significant role in using this spatial information.